Logical Aspects of Artificial Intelligence Introduction to Description Logics
Stephane´ Demri [email protected]
December 9th, 2019 What is in this part of the course?
Introduction to Description Logics and Temporal Logics for Multi-Agent Aystems
I Today: Introduction to description logics.
I 16/12/2019: Tableaux calculi and complexity.
I 06/01/2020: Introduction to temporal logics for multi-agent systems.
I 13/01/2020: 14h00–16h00 Exam on this part of the course (slides allowed). but also first-order logic, modal logics, knowledge representation, etc... 2 What can you expect to learn?
I Basics of description logic including ALC as well as ATL and variants.
I Tableaux for ALC, model-checking techniques for ATL-like logics.
I Complexity, decidability, expressive power results for logics dedicated to knowledge representation.
3 Background
1. Necessary background
I Basics of first-order logic.
I Basics of complexity theory.
2. Optional background
I Basics of modal logics, temporal logics
I Sequent-style proof systems.
I Basics of model-checking.
4 Course material
I Slides and exercises available on https://wikimpri.dptinfo.ens-cachan.fr/doku. php?id=cours:c-1-39 http://www.lsv.fr/˜demri/notes-de-cours.html
I Slides available after each lecture.
I Do not hesitate to contact me ([email protected]).
5 Material mainly based on the following documents
I F. Baader, I. Horrocks, C. Lutz and U. Sattler. An introduction to Description Logic. Cambridge University Press, 2017.
I Ivan Varzinczak’s slides (ESSLLI’18)
I Ulrike Sattler & Thomas Schneider’s slides (ESSLLI’15).
I S. Demri, V. Goranko, M. Lange. Temporal Logics in Computer Science. Cambridge University Press, 2016.
6 Other (online) ressources
I Description Logic Complexity Navigator by Evgeny Zolin. http://www.cs.man.ac.uk/˜ezolin/dl/
I Proceedings of the Description Logic Workshops http://dl.kr.org/workshops/
I See also the proceedings of the international conferences:
I Int. Joint Conference on Artificial Intelligence. (IJCAI) I European Conference on Artificial Intelligence. (ECAI) I Int. Conference on Principles of Knowledge Representation and Reasoning. (KR)
7 Plan of the lecture
I Knowledge representation.
I Basic description logic ALC.
I Several extensions of ALC.
I Relationships with first-order logic and modal logics.
I Exercises session.
8 Knowledge representation
9 DLs: where they come from
I First-order logic is not always the most natural language.
∀x ∃ y ∀z ((P(x, y) ∧ Q(y, z) ⇒ (¬Q(a, y) ∨ P(x, z))))
∀ x (Teacher(x) ⇔ Person(x)∧∃ y (Teaches(x, y)∧Course(y)))
I How to design user-friendly languages for knowledge representation ?
I Concept definition from Description Logics.
Teacher ≡ Person u ∃Teaches.Course
10 Reasoning about . . .
I Knowledge Epistemic Logics
I Rules and obligations Deontic Logics
I Programs Hoare Logics
I Time Temporal Logics but also many-valued logics, non-monotonic logics, team logics, separation logics, etc.
11 Ontologies
I Formal specification of some domain with concepts, objects, relationships between concepts, objects, etc.
I Backbone of ontologies includes: I taxonomy (classification of objects), I axioms (to constrain the models of the defined terms).
I Classification of medical terms: diseases, body parts, drugs, etc.
I Well-known ontologies: I Medical ontology SNOMED-CT formalised with description logic EL + +. I NCI Thesaurus (National Cancer Institute, USA). I Gene ontology (world largest source of information on the functions of genes).
I Free ontology editor Proteg´ e´ http://protege.stanford.edu/
12 The classical student ontology
I Natural language specification:
I Employed students are students and employees.
I Students are not taxpayers.
I Employed students are taxpayers.
I Employed students who are parents are not taxpayers.
I To work for is to be employed by.
I John is an employed student, John works for IBM.
I Classes/relations/individuals.
13 Main ingredients in formal ontologies
I Model of the world with classes within a domain, relationships between classes and instantiations of classes.
I Formal: abstract model of some domain with (mathematical) semantics and reasoning tasks.
I Classes or concepts: classes of objects with the domain of interest. (Employed student, Parent, Course)
I Relations or roles: relationships between concepts. (being employed by, sibling-of)
I Instances of classes and relations.
I John is an employed student. I Mary works for IBM.
14 Early KR formalisms
I Graphical formalisms easier to grasp and supposedly close to how knowledge is represented by human beings.
I Large variety of semantical networks.
I Often, lack of formal semantics (see tentatives with the knowledge representation system KL-ONE).
15 Why Description Logics?
I Formal languages for concepts, relations and instances.
I DLs have all one needs to formalise ontologies.
I Computational properties.
I Acceptable trade-off between expressivity and complexity. I Decidability and often tractability. I Implementation in tools of the main reasoning tasks.
I A remarkable suite of languages and tools. See e.g.
I OWL: Web Ontology Language. I Proteg´ e:´ ontology editor. I FaCT++: DL reasoner supporting OWL DL.
16 Description Logics and Knowledge representation
I Description is a subfield of Knowledge Representation, itself a subfield of Artificial Intelligence.
I Description Logic(s):
I a research field, I a family of knowledge representation languages, I a member of the family.
I Well-defined syntax with formal semantics, decision problems, algorithms, etc.
17 Basic description logic ALC
18 DLs: the core
I Concept language.
Person u ∃Teaches.Course
I Syntactic ingredients of the concept language:
I Concept names for sets of elements, e.g. Person. I Role names interpreted by binary relations between objects, e.g. EmployedBy. I Concept constructors to build complex concepts, e.g. ¬, u, t, ∃.
I Basic terminology stored in a TBox.
I Facts about individuals stored in an ABox.
19 Basic elements of the language
I Concept names.
def NC = {A1, A2,...}
Examples: Parent, Sister, Student
I Role names. def NR = {r1, r2,...} Examples: EmployedBy, MotherOf
I Individual names.
def NI = {a1, a2,...}
Examples: Mary, Alice, John
20 Boolean constructors & role restrictions
I Boolean constructors.
I Concept negation ¬ (class complement)
I Concept conjunction u (class intersection)
I Concept disjunction t (class union)
I Role restrictions. I Existential restriction ∃ (at least one related individual)
I Value restriction ∀ (all related individuals)
I Many more constructors exist, see forthcoming ALC extensions.
I For modal logicians, ¬, u, t, ∃, ∀ ∼ ¬, ∧, ∨, 3, 2
21 Complex concepts in ALC
I ALC: Attributive Concept Language with Complements.
I Complex concepts. C ::= > | ⊥ | A | ¬C | C u C | C t C | ∃r.C | ∀r.C,
where A ∈ NC and r ∈ NR.
I Examples of complex concepts:
I Student u ¬∃Pays.Tax I ∃MotherOf.(∃MotherOf.A)
I Syntax errors in Student t ∀¬Tax ∀∃MotherOf.Mary
def I C ⇒ D = ¬C t D.
22 Interpretation
I Concept/role/individual ∼ unary predicate/binary predicate/constant.
def I I I Interpretation I = (∆ , · )
I I ∆ : non-empty set (the domain).
I I · : interpretation function such that
AI ⊆ ∆I r I ⊆ ∆I × ∆I aI ∈ ∆I
I I A priori, ∆ is arbitrary and I can be viewed as a first-order model for unary and binary predicate symbols and constants.
23 Semantics for complex concepts
>I =def ∆I ⊥I =def ∅ (¬C)I =def ∆I \ CI I def I I (C1 t C2) = C1 ∪ C2 I def I I (C1 u C2) = C1 ∩ C2 (∃r.C)I =def {a ∈ ∆I | r I (a) ∩ CI 6= ∅} (∀r.C)I =def {a ∈ ∆I | r I (a) ⊆ CI }
(R(a) =def {b | (a, b) ∈ R})
I I In modal logic lingua, a ∈ C corresponds to I, a |= C.
24 Graphical representation
Teaches
C2 b Course Course Person Teaches
Teaches a C1 Person Course Teacher
I I ∆ = {a, b, C1, C2}. I I Teaches = {(a, C1), (a, C2), (b, b)}. I I I Person = {a, b}, Course = {C1, C2, b}. I I a ∈ (∀Teaches.Course) . 25 Concept satisfiability problem
I Concept satisfiability problem: Input: A (complex) concept C in ALC. Question: Is there an interpretation I = (∆I , ·I ) such that CI 6= ∅?
I This corresponds to the standard formulation for the satisfiability problem (in modal logics, temporal logics, etc.).
I The concept satisfiability problem for ALC is PSPACE-complete.
I ALC has the finite interpretation property: every satisfiable concept has an interpretation with a finite domain.
26 Statements
I Concept inclusion. Teachers are persons. Employed students are employees.
I Concept and role membership. Mary is a student. Alice is a teacher. Laura teaches the course “Automata Theory”.
I Statements are not concepts and express properties of concepts, roles and individuals.
27 General concept inclusion (GCI)
I Expressions of the form
C v D
are called general concept inclusion.
I Intuitive meaning:
I D subsumes C. I C is more specific than D.
I Example: Employee v ∃WorksFor.>.
def I I I Satisfaction relation: I |= C v D ⇔ C ⊆ D .
I C v D understood as a global statement about I.
28 Concept equivalence
I C v D and D v C abbreviated by
C ≡ D
called concept equivalence.
def I I I Satisfaction relation: I |= C ≡ D ⇔ C = D .
I Concept definition (A ∈ NC is a concept name) A ≡ C
I > ≡ (¬Student t Student).
29 Subsumption problem
I Subsumption problem: Input: A GCI C v D with C, D ∈ ALC. Question: Is it the case that for all interpretations I, we have I |= C v D?
I C v D is “not valid” iff C u ¬D is satisfiable.
I As coPSPACE =PSPACE, the subsumption problem for ALC is PSPACE-complete too.
30 Assertions
I Concept assertion: stating that an individual is an instance of a concept. a : C
def I I I Satisfaction relation: I |= a : C ⇔ a ∈ C .
I Role assertion: stating that two individuals are in a relation. (a, b): r def I I I I Satisfaction relation: I |= (a, b): r ⇔ (a , b ) ∈ r .
I Examples: I Alice : Student u ¬∃Pays.Tax. I (Laura, CNRS): WorksFor.
31 The validity problem
I Validity problem: Input: A statement α in ALC. Question: Is the case that for all interpretations I, we have I |= α?
I Validity of α is written |= α.
I Validity of > v C corresponds to the usual notion of validity for C, i.e. for all interpretations I = (∆I , ·I ), we have CI = ∆I .
I Examples of valid statements:
I |= ∀r.C u D v ∀r.C. I |= a : C t ¬C. I |= > v (¬(C u D) t (C t D)).
I The validity problem for ALC is PSPACE-complete.
32 What is a knowledge base (a.k.a. ontology) ?
I Terminological Box (TBox) T : finite collection of GCIs.
I I.e., a finite set of subsumption statements. I This provides definitions of concepts (a terminology).
I Assertional Box (ABox) A: finite collection of assertions.
I I.e., a finite set of concept and role assertions. I This provides a partial view on the interpretations and can be understood as a finite database.
I Knowledge base K is a pair (T , A).
I Knowledge bases are also called ontologies.
33 A knowledge base K?
I TBox T : Course v ¬Person Teacher v Person u ∃Teaches.Course ∃Teaches.> v Person Student v Person u ∃Attends.Course ∃Attends.> v Person
I ABox A: Mary : Person CS600 : Course Alice : Person u Teacher (Alice, CS600): Teaches (Mary, CS600): Attends
34 Consequences from knowledge bases
I I I Interpretation I = (∆ , · ), knowledge base K = (T , A). def I I |= A ⇔ for all α ∈ A, we have I |= α. def I I |= T ⇔ for all α ∈ T , we have I |= α. def I I |= K ⇔ I |= A and I |= T .
def I K |= α ⇔ for all interpretations I such that I |= K, we have I |= α.
I K? |= CS600 : ¬Person and K? |= Alice : Teacher.
35 Decision problems relatively to a knowlegde base
def I K is consistent ⇔ there is some I such that I |= K.
def I C is satisfiable with respect to K ⇔ there is I such that I |= K and CI 6= ∅.
def I C is subsumed by D with respect to K ⇔ K |= C v D.
def I C and D are equivalent with respect to K ⇔ T |= C ≡ D.
def I a is an instance of C with respect to K ⇔ K |= a : C.
I K |= C v D also written C vK D. K |= C ≡ D also written C ≡K D.
36 Relationships between reasoning problems
I C and D are equivalent w.r.t. K iff C is subsumed by D w.r.t. K and D is subsumed by C w.r.t. K.
I C vK D iff C u ¬D is not satisfiable w.r.t. K.
I C is satisfiable w.r.t. K iff C 6vK⊥.
I C is satisfiable w.r.t. K iff (T , A ∪ {b : C}) is consistent. (b is fresh)
I K |= a : C iff (T , A ∪ {a : ¬C}) is not consistent.
37 C is satisfiable w.r.t. K iff (T , A ∪ {b : C}) is consistent
I Suppose that C is satisfiable w.r.t. K. I I I There is I such that I |= K and C 6= ∅, say a ∈ C .
0 I0 def I Let I be the variant of I such that b = a.
0 I As b does not appear in K and C, we have I |= K.
0 I I0 I Furthermore, I |= b : C as C = C .
0 I Consequently, I |= (T , A ∪ {b : C}).
I Now, suppose that (T , A ∪ {b : C}) is consistent.
I There is I such that I |= T , I |= A and I |= b : C.
I I I Consequently, b ∈ C .
I I So, there is some I such that I |= K and C is non-empty.
38 Classification
I Deduce implicit knowledge from the explicitly represented knowledge.
I For all A, B in K, check whether A vK B.
I For all A in K, check whether A is satisfiable w.r.t. K. If not for some B, a modelling error is probable.
I For all a and C in K, check whether K |= a : C.
I Classifying a knowledge base K. 1. Check whether K is consistent, if yes, go 2. 2. For each pair A, B of concept names (plus >, ⊥), check whether K |= A v B. 3. For individual name a and concept C in K, check whether K |= a : C. leading to K’s inferred class hierarchy. 39 Complexity results for ALC
I Concept satisfiability and subsumption problems are PSPACE-complete. (no knowledge base involved)
I Knowlegde base consistency problem is EXPTIME-complete.
NP ⊆ PSPACE ⊆ EXPTIME ⊂ 2EXPTIME ⊂ N2EXPTIME
I Recall that C vK D iff (T , A ∪ {b : C u ¬D}) is not consistent.
40 Several extensions of ALC
41 Extensions: a feature of DLs
I Concepts/assertions in ALC have a limited expressive power.
I How to express simple arithmetical constraints such as “Alice teaches at least three courses”? I How to enforce constraints between roles? For instance, r I = (sI )−1 or r I ⊆ sI .
I The expressive power of ALC concepts can be characterised precisely, thanks to the notion of bisimulation (not presented today).
B Trade-off between the expressive power and the computational properties of the extensions.
I In the other direction: study of ALC fragments to reduce the complexity while preserving the expression of interesting properties, see e.g. EL or FL0.
42 Inverse roles Course v ¬Person Teacher v Person u ∃Teaches.Course ∃Teaches.> v Person Student v Person u ∃Attends.Course ∃Attends.> v Person Professor v Teacher Course v ∀TaughtBy.¬Professor
I Extending NR with inverse roles: − NR ∪ {r | r ∈ NR}
def I I − I def I −1 I Given I = (∆ , · ), (r ) = (r ) where R−1 =def {(b, a) | (a, b) ∈ R}
43 Elimination of the role name TaughtBy
I Back to the previous example. Professor v Teacher Course v ∀Teaches−.¬Professor
I Given a logic L, LI is defined as L except that inverse roles are added.
I Concept satisfiability for ALCI remains PSPACE-complete and knowledge consistency remains EXPTIME-complete.
44 Number restrictions
I How to express in ALC that a student attends to at least three courses? Student v ∃Attends.Course u A ∃Attends.Course u ¬A u B ∃Attends.Course u ¬A u ¬B
I Why isn’t it satisfactory ?
I How to express in ALC that a student attends to at most 10 courses?
I There is no concept C in ALC such that for all interpretations I, for all a ∈ ∆I , a ∈ CI iff card({b | (a, b) ∈ AttendsI }) ≥ 3
45 (Unqualified) number restriction
I Extending the concepts with number restrictions (≤ n r) and (≥ m r).
def I I I Given I = (∆ , · ),
(≤ n r)I =def {a ∈ ∆I | card({b | (a, b) ∈ r I }) ≤ n}
(≥ m r)I =def {a ∈ ∆I | card({b | (a, b) ∈ r I }) ≥ m}
I Given a logic L, LN is defined as L except that (unqualified) number restrictions are added.
I In ALCN , (≥ 3 Attends) u (≤ 10 Attends) does the job.
46 Qualified number restriction
I Generalising the number restrictions.
I Qualified number restrictions: (≤ n r · C), (≥ m r · C).
def I I I Given I = (∆ , · ),
def (≤ n r·C)I = {a ∈ ∆I | card({b | (a, b) ∈ r I and b ∈ CI }) ≤ n}
def (≥ m r·C)I = {a ∈ ∆I | card({b | (a, b) ∈ r I and b ∈ CI }) ≥ m}
I (∼ n r) = (∼ n r · >).
I Given a logic L, LQ is defined as L except that qualified number restrictions are added.
I Concept satisfiability for ALCIQ is PSPACE-complete and knowledge base consistency is EXPTIME-complete.
47 Naming individuals in complex concepts
I Concepts in ALC C ::= > | ⊥ | A | ¬C | C u C | C t C | ∃r.C | ∀r.C
I . . . but individual names a (as in the concept assertion a : C) are not concepts in ALC.
I How to express the concept “the courses taught by Alice” ? Course u ∃Teaches−.Mary
I Nominals in hybrid (modal) logics: propositional variables true in only one world of the domain.
I Nominals in DLs: individual names inside concept I def I descriptions, written {a}, where a ∈ NI with {a} = {a }. B Syntactic trick {·}: Course u ∃Teaches−.{Mary}.
48 Nominals in DLs
I Given a logic L, LO is defined as L except that nominals are added.
I Concept satisfiability for ALCOQ is PSPACE-complete and knowledge base consistency is EXPTIME-complete.
I . . . but concept satisfiability for ALCOI is EXPTIME-complete.
49 Role hierarchies
I ALC is not able to express complex role constraints such that a relation is included in another relation.
I Typically, the interpretation of Attends should include the interpretation of AttendsActively.
I Role inclusion axiom (RIA) of the form r v s with
I |= r v s ⇔def r I ⊆ sI
I Given a logic L, LH is defined as L except that role inclusion axioms are added (in the TBox).
I Concept satisfiability for ALCH is PSPACE-complete and knowledge base consistency is EXPTIME-complete.
50 Role value maps are concepts!
I A role value map is an atomic concept of the form r v s:
(r v s)I =def {a ∈ ∆I | r I (a) ⊆ sI (a)}
B Role value maps are local variants of role inclusion axioms (RIAs).
I The RIA r v s can be encoded by the GCI > v (r v s).
51 Transitive roles
I Many natural relations are transitive (AncestorOf, HasPart, etc.) but this cannot be expressed in ALCH.
I Transitivity axioms are of the form Trans(r):
I |= Trans(r) ⇔def r I ◦ r I ⊆ r I
I Given L, its extension with transitivity axioms in TBoxes is obtained by replacing ALC by S (new naming rule).
I Concept satisfiability for S is PSPACE-complete and EXPTIME-complete for knowledge base consistency.
I Other properties are included in knowledge bases such as reflexivity, irreflexivity, symmetry, functionality, . . .
52 A selection of complexity results
53 A queen logic SROIQ
I SROIQ agrees with the ontology language OWL 2 DL.
I SROIQ contains more than one might think from its name.
I Its knowledge bases contain a role box (RBox) to specify constraints about the interpretation of the role expressions.
I The set of roles R is made of role names r, its converses r − and the universal role U with UI =def ∆I × ∆I .
I New atomic concept ∃R.Self with (∃R.Self)I =def {a | (a, a) ∈ RI }
I Role assertions (a, b): ¬R are allowed in the ABox.
I + ingredients the name (nominals, qualified number restrictions, inverse). 54 Role axioms in the RBox
I Complex role inclusion axioms (CRIA) R1 ◦ · · · ◦ Rn v R.
B A regularity constraint is required on the set of CRIAs (unspecified here).
I Role axioms specifying disjointness, transitivity, reflexivity, irreflexivity, symmetry, asymmetry.
I The knowledge base consistency problem for SROIQ is N2EXPTIME-complete.
55 Semantic Web
I Semantic web:
I A vision of a computer-understandable Web. I Distributed knowledge and data in reusable form. I XML, RDF and OWL are part of the story.
I Principles towards a semantic Web of data
I Give a name to everything. I Relationships form a graph between the entities. I The names are addresses on the Web. I Provide a formal semantics so that knowledge is encoded in a machine interpretable way.
56 OWL based on description logics
I OWL: Web Ontology Language.
I Motivated by semantic web activities: add meaning to web content by annotating it with terms defined in ontologies.
I It is a World Wide Web (W3C) standard.
I OWL has an explicit formal semantics.
I Supported by tools and infrastructures such as development environments, reasoners and information systems.
I Based on description logics such as SROIQ.
57 DLs and ontology languages
I W3C’s OWL 2 is based on SROIQ.
I OWL 2 EL based on EL and OWL 2 QL based on DL-Lite.
I OWL was based on SHOIN .
I An OWL ontology is a mixed set containing TBox axioms and ABox assertions.
I More on complexity/scalability:
I OWL (SHOIN ) is NEXPTIME-complete. I OWL 2 EL is PTIME-complete.
58 OWL RDF/XML exchange syntax
Person u ∀Teaches.Course
59 OWL reasoners and Proteg´ e´
I OWL reasoners: implement decision procedures for consistency and ontology classification.
I Open-source ontology editor Proteg´ e.´
I Interaction with DL reasoners (FaCT++, Pellet, Racer) via the OWL API.
I Show results about ontology classification.
I Helpful to work with toy ontologies.
60 Relationships with other logics
61 Foreign language for modal/classical logicians
I Description logics can be understood as fragments of first-order logics.
I Similarly, reasoning tasks for description logics can be understood as decision problems for modal logics.
I This can be made precise and sometimes results for modal/first-oder logics can be used.
I Specificity of DLs: many fragments, many extensions and numerous original reasoning tasks (non-exhaustive presentation in this course).
62 Description logics and its shared history
I Late 80s: description logics developped as logical formalisms for semantics networks.
I In the 1990s: relationships with first-order logic, modal logics, PDL-like logics, etc. (PDL = Propositional Dynamic Logic)
I Logical basis for the Web Ontology Language OWL.
I Analogies between ontologies and databases lead also to relationships with query answering languages.
63 From concepts to first-order formulae
I I I Interpretations I = (∆ , · ) understood as first-order models.
I Translation of non-logical symbols.
Description logics First-order logic individual name a ∈ NI ≈ constant a concept name A ∈ NC ≈ unary predicate A role name r ∈ NR ≈ binary predicate r
I Translation of concepts, assertions and GCIs by internalising the DL semantics.
64 Example of translation
I A small TBox and its small ABox: ∃Attends.> v Person Teacher ≡ Person u ∃Teaches.Course Alice : Teacher
I Its translation in FOL:
∀ x (∃ y Attends(x, y) ⇒ Person(x)) ∧ ∀ x (Teacher(x) ⇔ (Person(x)∧∃ y(Teaches(x, y)∧Course(y)))) ∧ Teacher(Alice)
I Concepts can be understood as first-order formulae with one free variable.
65 Internalisation of ALC semantics
t(A, x) =def A(x) t(>, x) / t(⊥, x) =def > / ⊥ t(¬C, x) =def ¬t(C, x) def t(C1 u C2, x) = t(C1, x) ∧ t(C2, x) def t(C1 t C2, x) = t(C1, x) ∨ t(C2, x) t(∃r.C, x) =def ∃ y r(x, y) ∧ t(C, y) t(∀r.C, x) =def ∀ y r(x, y) ⇒ t(C, y) where y is a fresh variable. I I I I Given I = (∆ , · ), a ∈ C iff I, ρ[x ← a] |= t(C, x).
I Many-one reduction from ALC satisfiability problem to FOL satisfiability.
66 Translating KBs
t(C v D) =def ∀ x t(C, x) ⇒ t(D, x) t(C ≡ D) =def ∀ x t(C, x) ⇔ t(D, x) t(a : C) =def t(C, x)[a/x] t((a, b): r) =def r(a, b) where ϕ[a/x] (also written ϕ(a)) is the formula obtained from ϕ obtained by replacing the free occurrences of x by a.
I t(K) is defined as the conjunction below: ^ ^ t(α) ∧ t(α) α∈T α∈A
I I I Given I = (∆ , · ), I |= K iff I |= t(K).
67 Locating DLs within first-order logic fragments
I The definition of t(C, x) can be optimised to recycle variables and to use only two variables x0 and x1. def t(∃r.C, xi ) = ∃ x1−i r(xi , x1−i ) ∧ t(C, x1−i ) def t(∀r.C, xi ) = ∀ x1−i r(xi , x1−i ) ⇒ t(C, x1−i )
I FO2 (FOL restricted to two individual variables) satisfiability is NEXPTIME-complete.
I Actually, recycling of variables leads to the guarded fragment restricted to two variables GF2, whose satisfiability is EXPTIME-complete.
I Which additional DL features can be translated into FOL? into a decidable fragment of FOL?
68 More translations into FOL
t(∃r −.C, x) =def ∃ y r(y, x) ∧ t(C, y) t({a}, x) =def a = x t((≥ n r · C), x) =def ∃≥n y r(x, y) ∧ t(C, y) t(∃r.Self, x) =def r(x, x) t(r v s) =def ∀ x, y r(x, y) ⇒ s(x, y) def t(Trans(r)) = ∀x1, x2, x3 r(x1, x2) ∧ r(x2, x3) ⇒ r(x1, x3) def t(r1 ◦ · · · ◦ rn v r) = ∀x1, ··· , xn+1 r1(x1, x2) ∧ · · · ∧ rn(xn, xn+1) ⇒ r(x1, xn+1)
≥k def ^ ^ ∃ x ϕ(x) = ∃ x1,..., xk ( xi 6= xj ) ∧ ϕ(xi ) i6=j i
69 Modal logics in a nutshell
I Formulae: ϕ ::= p | ¬ϕ | ϕ ∧ ϕ | 3ϕ | 2ϕ.
I Kripke-style structures M = (W , R, V ):
I W : non-empty set of worlds. I R ⊆ W × W : accessibility relation. I V : PROP → P(W ): valuation. p |= 33p ∧ 33¬p ∧ 2¬p w q p, q q
I Satisfaction relation: def I M, w |= p ⇔ w ∈ V (p). def 0 0 0 I M, w |= 3ϕ ⇔ there is w s.t. (w, w ) ∈ R and M, w |= ϕ. def 0 0 0 I M, w |= 2ϕ ⇔ for all w s.t. (w, w ) ∈ R, M, w |= ϕ.
70 Ubiquity of modal logics
I Satisfiability problem: given a formula ϕ, are there M, w such that M, w |= ϕ?
I Plethora of modal logics depending on the frame conditions:
I Modal logic S5: R is an equivalence relation (or R = W × W ). I Modal logic K: R is arbitrary (or (W , R) is a finite tree). I Modal logic S4: R is reflexive and transitive.
I Epistemic/temporal logics can be viewed as modal logics with
I specific frame conditions (e.g., (W , R) is a tree), I multiple modalities (e.g., [r] for r ∈ NR), I modalities of arity > 1 (e.g., the until operator U). ϕUψ, ϕ ϕ ϕ ψ ...
71 Adding the universality modality and nominals
ϕ ::= A | a | ¬ϕ | ϕ ∧ ϕ | hriϕ | [r]ϕ | [U]ϕ | hUiϕ
I Models of the form M = (W , (Rr )r∈NR , V ) with associated modalities hri and [r].
def I University modality [U] such that M, w |= [U]ϕ ⇔ for all w 0, we have M, w 0 |= ϕ.
I Propositional variables are now denoted by A, B to prepare the encoding.
I Nominals in hybrid modal logics are propositional variables holding for a unique world, here represented by a, b.
I Satisfiability problem for this logic is EXPTIME-complete.
72 Internalisation of the semantics t(A) =def A t(>) / t(⊥) =def > / ⊥ t(¬C) =def ¬t(C) def t(C1 u C2) = t(C1) ∧ t(C2) def t(C1 t C2) = t(C1) ∨ t(C2) t(∃r.C) =def hrit(C) t(∀r.C) =def [r]t(C)
I I I I = (∆ , · ) 7→ MI = (W , (Rr )r∈NR , V ) def I I I W = ∆ and Rr = r for all r. def I def I I V (A) = A for all A and V (a) = {a } for all a.
I I I I Given I = (∆ , · ), a ∈ C iff MI , a |= t(C).
I Many-one reduction from ALC satisfiability problem to modal satisfiability.
73 Translating KBs
t(C v D) =def [U](t(C) ⇒ t(D)) t(C ≡ D) =def [U](t(C) ⇔ t(D)) t(a : C) =def hUi(a ∧ t(C)) t((a, b): r) =def hUi(a ∧ hrib)
I t(K) is defined as the conjunction below: ^ ^ t(α) ∧ t(α) α∈T α∈A
I I I I Given I = (∆ , · ), I |= K iff MI , a |= t(K) for all a ∈ ∆ .
74 Low hanging fruits
+ I Let ML (resp. ML ) be the multimodal logic into which ALC concepts (resp. knowledge bases) are translated.
I ML satisfiability problem is PSPACE-complete, it has the finite model property and the tree model property.
+ I ML satisfiability problem is EXPTIME-complete, it has the finite model property and the tree model property.
I ALC satisfiability problem is PSPACE-complete and it has the finite interpretation property.
I ALC knowledge base consistency problem is EXPTIME-complete and it has the finite tree-shaped interpretation property.
75 Recapitulation
76 Recapitulation: concept and role constructors
Name Syntax Semantics Top > ∆I Bottom ⊥ ∅ Conjunction C u D CI ∩ DI Disjunction C t D CI ∪ DI Negation ¬C ∆I \ CI Existential restr. ∃r.C {a ∈ ∆I | r I (a) ∩ CI 6= ∅} Value restr. ∀r.C {a ∈ ∆I | r I (a) ⊆ CI } Unqual. nb. restr. (≤ n r) {a ∈ ∆I | card({b | (a, b) ∈ r I } ≤ n)} Qual. nb. restr. (≤ n r · C) {a ∈ ∆I | card({b ∈ CI | (a, b) ∈ r I } ≤ n)} Nominal {a} {aI } Role value map r v s {a ∈ ∆I | r I (a) ⊆ sI (a)} Inverse role r − {(b, a) | (a, b) ∈ r I } Role composition r ◦ s {(a, b) | ∃ a0 (a, a0) ∈ r I and (a0, b) ∈ sI }
77 Recapitulation: Terminological and assertional axioms
Name Syntax Semantics General inclusion axiom C v D CI ⊆ DI Concept definition A ≡ C AI = CI Role inclusion r v s r I ⊆ sI Role transitivity Trans(r) r I is transitive Concept assertion a : C aI ∈ CI Role assertion (a, b): r (aI , bI ) ∈ r I
78 Conclusion
I Lecture 1 (today): Introduction to description logics
I Getting familiar with DL terminology.
I Playing with formulae and decision problems.
I Lecture 2: Tableaux proof systems and complexity .
I Complete calculi for ALC and variants.
I Complexity results.
I Undecidability.
I Lecture 3: Introduction to temporal logics for multi-agents systems.
79